TIGIT and PD-1 Mark Intratumoral T Cells with Reduced Effector Function in B-cell Non-Hodgkin Lymphoma

T-cell activation is initiated upon antigen recognition by the T-cell receptor (TCR) and is further potentiated by activation of costimulatory receptors (1). This is counteracted by coinhibitory receptors such as CTLA-4 and PD-1, which are transiently induced upon TCR activation to balance acute immune responses. In chronic infection and cancer, high expression of coinhibitory receptors persists, leading to impaired T-cell function (2). Consequently, immune checkpoint blockade targeting coinhibitory receptors such as PD-1 has emerged as a promising immunotherapeutic approach (3). However, although PD-1/PD-L1 blockade as monotherapy has been successful in relapsed/refractory (R/R) Hodgkin lymphoma with objective response rates (ORR) of 65% to 87% (4–7), the benefits have overall been modest in non-Hodgkin lymphoma (NHL), demonstrated by ORR of 40% in R/R follicular lymphoma (FL; ref. 8), 36% in R/R diffuse large B-cell lymphoma (DLBCL; ref. 8) and no responders in relapsed chronic lymphocytic leukemia (CLL; ref. 9), despite frequent expression of PD-1 and PD-L1 (10). Data on PD-1 blockade in mantle cell lymphoma (MCL) remain incomplete. As progression of T-cell exhaustion is linked to expression of increased numbers of coinhibitory receptors (2), checkpoint coblockade may be necessary to achieve optimal antitumor T-cell responses. However, although PD-1 expression has been studied in NHL (10), expression of other immune checkpoint receptors is less well characterized. Investigation of coinhibitory receptor expression patterns is warranted to determine relevant targets for checkpoint blockade.

TIGIT (T-cell immunoglobulin and ITIM domain) is a coinhibitory receptor that can be expressed by effector T cells, natural killer (NK) cells, T regulatory cells (Treg) and T follicular helper (T_FH) cells (11–14). TIGIT has gained attention as a potential therapeutic target in cancer due to its frequent expression on tumor-infiltrating T cells and its association with CD8⁺ T-cell exhaustion (15–17). The TIGIT ligands, CD155 and CD112, can be expressed by different cell types, including antigen-presenting cells and tumor cells (11, 18, 19). We previously identified TIGIT as a potential target for checkpoint blockade in FL by demonstrating that intratumoral CD8⁺ T cells with dysfunctional TCR signaling were identified by TIGIT expression (20). Here, multicolor flow cytometry was used to characterize the landscape of coinhibitory receptor expression in distinct T-cell subsets from DLBCL, MCL, FL, CLL, and marginal zone lymphoma (MZL). Our aim was to identify relevant checkpoint receptors for clinical investigation as targets for checkpoint blockade in NHL. We examined the numbers of TIGIT and PD-1–positive intratumoral T cells, correlated TIGIT and PD-1 expression with the T cells' capacity to produce cytokines, and also report expression of the TIGIT and PD-1 ligands in the tumor microenvironment.

Samples were obtained with informed written consent in accordance with the Declaration of Helsinki and with approval from the Regional Committee for Medical and Health Research Ethics. Tumor biopsies were obtained from patients with FL (n = 19), DLBCL (n = 19), MCL (n = 11), CLL (n = 7), and MZL (n = 2) at the Norwegian Radium Hospital, Oslo, Norway; clinical characteristics are described in Supplementary Table S1. DLBCL samples included germinal center B-cell (GCB; n = 4) and non-GCB (n = 15) subtypes. Tonsils were obtained from patients (n = 19) undergoing tonsillectomy at Agroklinikken. Samples were processed to single-cell suspensions by mincing and cryopreserved in liquid nitrogen. A number of the FL (n = 14) and tonsil (n = 10) specimens had been included in our previous study (20).

Flow cytometry analysis was performed as previously described (20). Single cells were stained with Alexa Fluor 594 dye (Thermo Fisher), to exclude dead cells from analysis, and then fixed in paraformaldehyde (PFA; 1.6%). Fixed cells were stained with the following antibodies: CD3-Pacific Blue (clone UCHT1), CCR7-PE (150503), CXCR5-Ax488 (RF8B2), CD20-APCH7 (L27), PDL1-APC (MIH1), PDL2-APC (MIH18), and IFNγ-PE (4S.B3) from BD Biosciences; TIGIT-APC (MBSA43), LAG3-PeCy7 (3DS223H), TNFα-Ax488 (MAb11), and IL2-PeCy7 (MQ1-17H12) from eBioscience; and CD4-Ax700 (RPA-T4), CD8-Bv785 (RPA-T8), CD45RA-Bv510 (HI100), PD1-Bv650 (EH12.2H7), TIM3-APC (F38-2E2), BTLA-APC (MIH26), CD244-PerCPCy5.5 (C1.7), CD160-PeCy7 (BY55), LAIR1-PerCPCy5.5 (NKTA255), CD155-PE (SKII.4), and CD112-PeCy7 (TX31) from BioLegend. Brilliant Stain Buffer (BD Biosciences) was used as staining buffer. Data were acquired on LSR II (BD Biosciences) and analyzed using Cytobank (https://www.cytobank.org/).

Serial sections of cryopreserved tissue were stained with antibodies for CD155 (L95) and CD112 (L14) as previously described (20), in addition to PD-L1 (405.9A11) and CD68 (KP1).

T cells from NHL tumors were enriched by depletion of CD19⁺ B cells using Dynabeads CD19 Pan B (Thermo Fisher) according to the manufacturer's protocol. Cytokine production was then activated for 6 hours using Dynabeads Human T-Activator CD3/CD28 (Thermo Fisher) in a 1:1 bead-to-cell ratio, with GolgiPlug (BD Biosciences) present for the last 4 hours. Cells were fixed in 1.6% PFA to stop activity, followed by centrifugation and permeabilization in >90% ice-cold methanol. Samples were stored at −80°C before staining with antibodies and flow cytometry acquisition.

The lymphoma cell line SU-DHL-4 was a kind gift from Dr. Staudt, NCI, Bethesda, MD. The cells were sustained in RPMI-1640 (PAA Laboratories) supplemented with 10% fetal calf serum (FCS) and streptomycin/penicillin (PAA Laboratories) and were Mycoplasma tested with VenorGeM Classic (Minerva BioLabs). Cell line authentication was determined by PCR-single-locus technology (Promega, Powerplex 21 PCR kit), performed by Eurofins.

The full-length sequence encoding human PD-L1 was ordered as a synthetic DNA to Eurofins MWG and subcloned into pENTR (Thermo Fisher Scientific) using unique NotI/XhoI sites added before the start and after the STOP codons, respectively. The entry vector was then used to subclone the insert into a retroviral expression vector (pMP71) made compatible with the Gateway system as previously described (21). The human CD155 coding sequence was extracted from the commercial plasmid pCMV3-C-HA-CD155 (#HG10109-CY, Sino Biological) by PCR to remove the HA tag and to create the original isoform-1 using the following primers: 5′-CAC CAT GGC CCG AGC CAT G-3′ and 5′-GAA TTC ACC TTG TGC CCT CTG TCT GT-3′. The amplicon was subsequently subcloned into pENTR-D/TOPO vector (Thermo Fisher Scientific) and sequence verified before recombination into pMP71. Both expression constructs were deposited at Addgene (pMP71- hCD155, #118630 and pMP71-hPDL-1, #118631).

Viral particles were produced as previously described (21). Briefly, Hek-Platinum cells were transfected using Fugene-6 (Roche) with retroviral packaging vectors and the expression vectors. After 24 hours of incubation at 37°C, medium was replaced with DMEM containing 1% FCS, and cells were incubated at 32°C. Viral supernatants were harvested at 48 and 72 hours after transfection. Spinoculation of SU-DHL-4 cells was performed with 1 volume of retroviral supernatant in a 12-well culture nontreated plate (Nunc) precoated with retronectin (20 mg/mL, Takara Bio.). Cells were harvested with PBS-EDTA (0.5 mmol/L), but cultured for at least 2 weeks before they were used in experiments.

Characterization of coinhibitory receptor expression revealed that TIGIT and PD-1 were expressed at higher frequency than all other receptors investigated (Fig. 1A). TIGIT was expressed at increased frequency on T cells from NHL tumors as compared with tonsillar T cells from healthy donors (Fig. 1B). Gene-expression analysis has previously demonstrated upregulation of TIGIT and PD-1 in FL and DLBCL as compared with normal controls (22). In FL, high expression of both TIGIT and PD-1 in CD4⁺ T cells correlated with advanced disease stage (Supplementary Fig. S1A–S1D). Our results further showed that TIGIT and PD-1 surface expression varied between T-cell subsets.

Although few naïve T cells expressed TIGIT or PD-1, the majority of CD8⁺ and CD4⁺ T effector memory (T_EM) cells were positive for the two coinhibitory receptors (Supplementary Figs. S2 and S3; Fig. 2A and B). T-cell distribution was skewed toward T_EM cells, the main subset among CD8⁺ as well as CD4⁺ intratumoral T cells across all NHLs investigated (Supplementary Fig. S4A and S4B). Tumor-infiltrating CD8⁺ and CD4⁺ T_EM cells had TIGIT median expression ranging from 83% to 95% and 85% to 93%, respectively (Fig. 2A and B). PD-1 median expression ranged from 81% to 85% in CD8⁺ T_EM cells, and 70% to 75% in CD4⁺ T_EM cells (Fig. 2A and B). CD8⁺ T_EM cells in NHL, independent of type of lymphoma, expressed TIGIT and PD-1 at increased frequencies compared with tonsillar T cells. The percentage of TIGIT⁺ cells was also significantly increased in NHL CD4⁺ T_EM cells as compared with the tonsillar counterpart (Fig. 2A and B). In contrast, CD244 was expressed at lower and variable frequencies among CD8⁺ T_EM cells, also within the same type of NHL, whereas BTLA, TIM-3, LAG-3, CD160, and LAIR-1 were expressed by on average <20% of CD8⁺ and CD4⁺ T_EM cells (Fig. 2A and B; Supplementary Figs. S2 and S3).

As coexpression of TIGIT and PD-1 has been associated with T-cell dysfunction in cancer (15), we next investigated coexpression of these two receptors in NHL (Fig. 2C). TIGIT and PD-1 were coexpressed by the majority of T_EM cells. On average, 78% to 83% of CD8⁺ T_EM cells and 69% to 79% of CD4⁺ T_EM cells coexpressed the two coinhibitory receptors, across the different NHLs (Fig. 2D). A significantly higher degree of coexpression was observed in CD8⁺ T_EM cells from all lymphoma types investigated as compared with the tonsillar counterpart (Fig. 2D). For CD4⁺ T_EM cells, TIGIT and PD-1 coexpression was significantly increased in FL and CLL/MZL (Fig. 2D).

To benefit from checkpoint blockade, patients must previously have developed tumor-specific T cells. Hence, if the population of intratumoral TIGIT⁺PD-1⁺ T_EM cells found in NHL is enriched for exhausted, neoantigen-experienced T cells, targeting these receptors by checkpoint blockade might be efficient in restoring antitumor reactivity. Support for this hypothesis comes from emerging evidence that tumor neoantigens can be expressed in NHL. In MCL, neoantigenic mutations presented by MHC were exclusively derived from the lymphoma immunoglobulins and could induce T-cell responses (23). Furthermore, analysis of mutations, transcriptional profiles and single-cell TCR sequences in melanoma tumor samples obtained before and during treatment, revealed expansion of T-cell clones and putative selection against neoantigenic mutations in patients who responded to PD-1 blockade by nivolumab (24).

We hypothesized that intratumoral T-cell function correlated with expression of TIGIT and PD-1. To test this, cytokine production was measured in relation to TIGIT and PD-1 expression in CD4⁺ and CD8⁺ T cells from NHL. T cells expressing either TIGIT or PD-1 produced little IFNγ, TNFα, and IL2 compared with cells negative for the receptors (Fig. 3A). Among CD8⁺ T cells, intracellular TNFα and IL2 were significantly reduced in PD-1⁺TIGIT⁺ and PD-1⁺TIGIT⁻ cells compared with PD-1⁻TIGIT⁻ cells; the same trend was observed for IFNγ (Fig. 3B). Among CD4⁺ T cells, the capacity to produce TNFα and IL2 was significantly reduced in PD-1⁺TIGIT⁺ cells as compared with PD-1⁻TIGIT⁻ or PD-1⁺TIGIT⁻ cells, whereas IFNγ production was lower in PD-1⁺ TIGIT⁺ and PD-1⁻TIGIT⁺ cells as compared with PD-1⁺TIGIT⁻ cells (Fig. 3B). Collectively, these results indicate that TIGIT and PD-1 contribute to suppressed T-cell effector function. In line with this, intratumoral T cells cultured in vitro for 48 hours before activation of cytokine production had improved capacity to produce IL2 in TIGIT⁺ CD8⁺ T cells and TNFα in PD-1⁺ CD4⁺ T cells as compared with day 0 cultured cells (Supplementary Fig. S5). This suggested that the impaired effector function of TIGIT⁺ and PD-1⁺ T cells can be restored upon disruption of the in vivo immunosuppressive tumor microenvironment.

Coinhibitory receptors need to be ligated to exert their suppressive functions. We next investigated the expression of the TIGIT and PD-1 cognate ligands, CD155/CD112 (18) and PD-L1/PD-L2 (25), respectively. IHC analysis revealed expression of CD155 and CD112 on endothelial cells, including high endothelial venules and sinusoid endothelium, as well as on follicular dendritic cells (FDC) in DLBCL, MCL, and CLL (Fig. 4A; Supplementary Table S2). This is similar to what we observed in FL tumors (20). CD155 expression could also be observed in tumor cells from MCL and DLBCL (Fig. 4B). In contrast, PD-L1 demonstrated a different staining pattern and was expressed by intratumoral macrophages as revealed by morphology and CD68 expression (Fig. 4A; Supplementary Table S2). Flow cytometry analysis confirmed CD155 and CD112 expression in tumor cells in 7 and 9 out of 18 DLBCL cases, and in 3 and 2 out of 10 MCL cases, respectively (Fig. 4C). In contrast, tumor cells were negative for CD112 and CD155 in all cases of FL and CLL/MZL (Fig. 4C). PD-L1/PD-L2 was only expressed by tumor cells in 2 of the DLBCL cases (Fig. 4C). These results suggest that TIGIT and PD-1 are likely to inhibit T-cell antitumor activity through interaction with the tumor microenvironment in FL, MCL, and CLL/MZL, as well as by ligand-expressing tumor cells in DLBCL.

As TIGIT competes for ligand binding with the costimulatory receptor CD226 (26), we also investigated expression of this receptor (Supplementary Fig. S6A). Our results revealed that CD8⁺ T_EM cells, which displayed the highest expression of TIGIT (Supplementary Fig. S2), were CD226^low (<10% median expression across NHL; Supplementary Fig. S6B). In contrast, CD226 was frequently expressed in CD4⁺ T cells, including FL T_FH cells as previously reported (Supplementary Fig. S6C; ref. 20). Together, this indicated that low expression of CD226 may play a role in TIGIT-mediated inhibition of CD8⁺ T cells in NHL.

Whereas PD-1 blocks signaling events downstream of the TCR by recruiting the protein tyrosine phosphatases SHP1 and SHP2 (27), TIGIT can inhibit effector function by recruiting the inositol 5-phosphatase SHIP1 (28). This, in context with our discovery that the majority of intratumoral CD8⁺ and CD4⁺ T_EM cells coexpress TIGIT and PD-1, and the finding that TIGIT⁺PD-1⁺ T cells are poor producers of proinflammatory cytokines, indicates that dual blockade of these receptors might enable increased T-cell activity and tumor killing in NHL, potentially by recruitment of different phosphatases. Although not yet explored in lymphoma, a number of publications demonstrate synergistic activity of PD-1 and TIGIT coblockade in preclinical cancer models. Combined blockade of the two receptors resulted in complete responses in tumor mouse models of breast and colorectal cancers, whereas blocking only one receptor had little effect (15, 16, 29). Furthermore, an anti-mouse TIGIT-blocking antibody increased survival of glioblastoma bearing mice when administered in combination with PD-1 blockade in vivo, and the cured mice developed long-term immunologic memory protection upon rechallenge with tumor cells (29). To study the effect of TIGIT and PD-1 ligands in vitro, we introduced expression of PD-L1 and CD155 in the B-cell lymphoma cell line SU-DHL-4 (Supplementary Fig. S7A). However, FL T cells cocultured with the original ligand-negative cell line had higher cytokine production as compared with T cells cultured in medium alone, and this difference was greater than between T cells cocultured in the presence or absence of ligand-positive cells (Supplementary Fig. S7B). Despite limited functional effect of TIGIT blockade in vitro, PD-1 and TIGIT coblockade has potent antitumor activity in a mouse model (29). Furthermore, TIGIT may influence several types of immune cells. In addition to inhibiting effector T cells (12), TIGIT has been shown to suppress immune responses by dampening NK cell activity (28) and by inducing immunoregulatory dendritic cells (11). Hence, blocking the TIGIT pathway in these cells may also be necessary for efficient immunotherapy responses. Moreover, TIGIT is required for the B-cell helper function of circulating T_FH cells (30), suggesting that TIGIT might have different roles in distinct T-cell subsets. Although T_FH cells are typically absent in DLBCL, CLL, MCL, and MZL (Supplementary Fig. S4), TIGIT⁺ T_FH cells are frequently found in FL (Supplementary Figs. S3 and S4; ref. 20). Thus, TIGIT blockade might also promote antitumor responses by reducing the tumor supporting effect of T_FH cells in FL. Furthermore, Tregs are considered a suppressive barrier to potent antitumor responses in NHL, and we have shown that Tregs from FL tumors express TIGIT (20). In contrast to the unresponsive phenotype of effector T cells, TIGIT⁺ Tregs are more potent suppressors of proinflammatory immune responses as compared with TIGIT⁻ Tregs (13). Hence, immunotherapy using TIGIT-blocking antibodies might promote immune responses in different ways, including restoring antitumor potential of effector T cells or reducing Treg immunosuppression.

Taken together, our results demonstrate frequent expression of TIGIT and PD-1 in exhausted T cells from NHL tumors, and the presence of ligand-positive cells in the tumor microenvironment (Supplementary Fig. S8). Overall, this study provides an overview of the coinhibitory receptor landscape in NHL and serves as a reference map of checkpoint receptors relevant to investigate in a clinical setting. Preclinical models demonstrate complete responses following coblockade of TIGIT and PD-1 in solid cancer (15, 16, 29). Consistent with this, our findings indicate the value of combinatorial blockade of TIGIT and PD-1 in NHL and highlight the importance of characterizing the tumor microenvironment to understand mechanisms of immune escape.

R. Levy is a consultant/advisory board member for Five Prime, Immune Design, Beigene, Teneobio, Sutro, Checkmate, Dragonfly, Apexigen, and Forty Seven. A. Kolstad reports receiving commercial research funding from Nordic Nanovector and Merck and is a consultant/advisory board member for Nordic Nanovector and Celgene. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S.E. Josefsson, E.B. Smeland, K. Huse, J.H. Myklebust

Development of methodology: S.E. Josefsson, K. Beiske, J.H. Myklebust

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.E. Josefsson, K. Beiske, Y.N. Blaker, H. Holte, B. Østenstad, E. Kimby, A. Kolstad

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.E. Josefsson, K. Beiske, E.B. Smeland, K. Huse, J.H. Myklebust

Writing, review, and/or revision of the manuscript: S.E. Josefsson, Y.N. Blaker, H. Holte, E.B. Smeland, R. Levy, A. Kolstad, K. Huse, J.H. Myklebust

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S.E. Josefsson, M.S. Førsund, H. Köksal, A. Kolstad

Study supervision: E.B. Smeland, K. Huse, J.H. Myklebust

Other (plasmid constructs design, cloning, and expression): S. Wälchli

Other (performed genetic engineering): B. Bai

Other (provided patient samples and clinical data): Y.N. Blaker

We thank Daniela Pende for kindly providing monoclonal antibodies L95 and L14 and Vera Hilden for excellent technical assistance. This work was supported by the Research Council of Norway (FRIMEDBIO 230817/F20; S.E. Josefsson), Centre of Excellence (Centre for Cancer Biomedicine; J.H. Myklebust and E.B. Smeland), and the Norwegian Cancer Society (162948; K. Huse, 163151; E.B. Smeland, and 162844; J.H. Myklebust).